New method for positioning detection of coronary stent radiopaque holes

Abstract

Since a coronary stent is sleeved on a rotating support bar and the radiopaque holes are curved, it is necessary to control radiopaque holes face upwards during the automatic assembly of coronary stent radiopaque markers. In this paper, a new method is proposed for positioning detection of radiopaque holes based on rough positioning and precision positioning. In rough positioning, support vector machine (SVM) classification is used to learn the histograms of oriented gradients (HOG) in order to identify the presence or absence of radiopaque holes in the camera's vision. In precision positioning, the deviation angle of radiopaque holes is determined by HSV color space threshold segmentation and the fit ellipse of sub-pixel edges. The experimental results show that the proposed method can accurately and effectively detect the positive position of coronary stent radiopaque holes.